Comments on: An Exercise in Elementary Statistics: Application to Italian Electoral Polls

By: More Statistics… « Gravitas Free Zone Weblog

More Statistics… « Gravitas Free Zone Weblog — Sat, 29 Mar 2008 00:38:13 +0000

[...] Gravitas Free Zone Weblog floating in the web « An Exercise in Elementary Statistics: Application to Italian Electoral Polls [...]

By: sinigagl

sinigagl — Fri, 28 Mar 2008 16:48:32 +0000

Roberto, con l’aggiornamento di domani metterò un link a questo post: facciamo scambio? Ciao, Paolo

Certo, a dopo.
Roberto

By: sinigagl

sinigagl — Fri, 28 Mar 2008 15:59:10 +0000

I would like you to see http://www.politiche08.org which predicts 156 for PdL and 159 for all the others at Senato. Tomorrow now regional polls will be added. I’m sorry but it’s in italian!

I have seen it - good job! As I commented above, I had started introducing the regional polls myself. At present I have 158 for PdL, but I still miss some poll, and a global check. No time right now, will update later.

And, of course, I’m Italian myself. It’s a bit awkward to use English here, since all comments on this post are indeed by Italians for the moment, but I would like to keep it readable by foreigners. I reply in Italian to comments in Italian, though.
Cheers,
Roberto

By: Roberto

Roberto — Fri, 28 Mar 2008 15:04:19 +0000

Caminadella:
Some regional polls were indeed published in the last few days, limited to the most influential regions. One must understand, each regional poll has roughly the same number of interviews as a national one, so it will cost the same…
I did an exercise over lunch, including some of them. Still need to add a few ones (Piemonte and Campania, for instance), and do a final check. Will post an update here. But true, small changes can make a big difference, and be not even visible locally - see below for an example.

Dorigo:
You indeed choose to live in one of the craziest example of “third party porcellum strategy”.
As I mentioned, I just included regional polls for Veneto in the model. No changes from before (PdL 15, Pd 9). But UDC is locally estimated at 7.9 % (was 7.2 % with simple scaling from 2006). If it would reach 8 %, the results will be: (PdL 14, PD 8, UDC 2). So, if one wants to vote AGAINST PdL, the best bet would be to vote UDC, irrespectively of his political position. Note that the amount of votes needed for the swing are less than three thousands, and corresponds to ONE SINGLE PERSON in the regional poll. I think porcellum is not an underserved label after all.

By: Caminadella

Caminadella — Fri, 28 Mar 2008 14:11:21 +0000

On a related note, it’s quite unbelievable that pollsters aren’t doing more regional polls. Small voting differences in Lazio or Piemonte can change the balance much more than big differences in the aggregate data…

By: dorigo

dorigo — Fri, 28 Mar 2008 06:51:46 +0000

Ciao,
actually I vote in Veneto
I understand little of this porcellum thing though. Will have to study a bit.
Cheers,
T.

By: dorigo

dorigo — Thu, 27 Mar 2008 23:00:22 +0000

Honored. And thank you for the very nice analysis. Do you have an idea where I can find some discussion of the weird senate method of assigning seats and an analysis of what someone willing to favor PD or PDL should really vote for (such as voting for a third party if one of the two is surely in the lead in the region) ?

Cheers,
T.

Thanks to appreciate. The analysis made by Brusco in noiseFromAmeriKa is very informative on the “porcellum” method for the senate. A discussion on the possibilities for Italians abroad (in Europe) did appear in Scandinaria recently, I’ve commented on it in a previous post. Third party voting is a clear option in Emilia and Toscana (SA to subtract seats to PdL).
In other regions I think the situation is less clear.
Cheers,
Roberto